SOURCES OF THE ORES

Both the geologic setting and isotopic signatures of the REE ore minerals suggest a crustal source for REE. The formations underlying the Proterozoic Bayan Obo Group consist of Archean and Early Proterozoic granitic and migmatitic gneiss intruded by pegmatitic dikes and stocks. According to the Institute of Geochemistry (1988, p. 505), the Early Proterozoic granitic rocks and the widespread Early Proterozoic pegmatites contain relatively high contents of RE2O3, which for the granitic rocks range from 800 to 1,308 ppm, notably higher than the average of about 290 ppm RE2O3 for granitic rocks elsewhere. The granite-pegmatites in the area of Hayehhutung locally contain more than 10 volume percent allanites, some of which occur as round or lenticular bodies exceptionally high in LREE's (Zhao Zhenhua, oral commun., 1993). Some of the pegmatitic rocks contain Nb-Ta minerals. The only direct evidence we found with regard to RE2O3 sources was the occurrence of above-normal concentrations of RE2O3 in detrital apatite (see sample 7B25-1, table 5) from coarse-grained H8 dolostone marble.

The abundance of allanite in Early Proterozoic rocks in the Bayan Obo area, and the widespread low- to minable-grade monazite disseminated in ores throughout the 16 km by 2-3 km extent of the mine region, led investigators of the Institute of Geochemistry (1988) to favor a syngenetic origin for the Bayan Obo REE ores. This interpretation is not supported by our studies (Chao and others, 1992, 1993; Wang and others, 1994; this paper).

Pb and Nd isotopes from Bayan Obo ore and gangue minerals, analyzed by Wang and others (1994), indicate a crustal origin for the REE's at Bayan Obo. Typically, the 206Pb/204Pb ratios are low and the 208Pb/204Pb ratios are high for the REE-bearing and REE minerals. In the same suite of REE minerals, U is strongly depleted with respect to Th, and monazite, bastnaesite, REE-bearing apatite, and aegirine augite have high negative epsilon Nd values (-16 to -19). These isotopic signatures clearly indicate a crustal source for Bayan Obo REE ores.

TECTONIC SETTING

Caledonian subduction, related to REE mineralization at Bayan Obo, has been discussed in parts I and II (figs. 4 and 34).

MECHANISM FOR CONCENTRATING THE ORES IN SOURCE RESERVOIRS

We propose that the REE's and Fe were scavenged and concentrated from the underlying crust by acidic hydrothermal solutions activated by Caledonian subduction. In addition, repeated fracturing and shearing during the Paleozoic Era produced the channels for the hydrothermal solutions to bring the REE's and Fe from deep crustal source areas and deposit them mostly in the H8 host marble. On the basis of the experimental studies of Hemley and others (1992), we suppose that such long-distance travel by hydrothermal solutions enriched with scavenged REE's and Fe is feasible and probable.

MINERAL PARAGENETIC SEQUENCE AND GEOCHEMICAL EVOLUTION OF THE BAYAN OBO ORE DEPOSIT

The geochemical evolution of the Bayan Obo ore deposit can be deduced from the paragenetic relations of the hydrothermal minerals (table 13). REE's evidently were the first ore elements introduced, at about 555 Ma, along with P and F. The solutions containing these elements had random spatial and time distribution and produced disseminated ores containing monazite or bastnaesite or mixtures of these two major REE minerals. These disseminated REE ores were followed by the formation of REE banded ores, and later by Fe2O3 (hematite) ores, which together with martite and magnetite formed the massive Fe cores of the Main and East Orebodies.

During the peak periods of ore formation, REE's, Fe, P (to form monazite and apatite), Na, Si (to form aegirine augite), F (to form bastnaesite and fluorite), and Ba (to form barite) were brought in by different hydrothermal solutions in order to produce the wide variety of banded ores. Na, K, and Si (to form alkali amphiboles) were also introduced during the peak and waning phases of REE mineralization, but deposited in different parts of the orebodies than the above assemblage. During this period, U and S were uniquely missing as geochemical components.

During late-stage hydrothermal activities, Nb took on increasing importance. Nb accompanied and followed Fe mineralization, and was accompanied and followed by minor S activity (to form gangue sulfides). The last stage of hydrothermal activity consisted of Ba and K metasomatism, which was widespread in the H9 biotite schist (Drew and others, 1990).

Geochemically, the Bayan Obo ore deposit was enriched by ore metals in the sequence REE's, Fe, and Nb; and the gangue elements P, F, and Ba. Bayan Obo is characterized by the scarcity or lack of U, S, Ta, Al, the precious metals Au and Ag, and base metals Cu, Pb, and Zn.

MINERAL ASSEMBLAGES AND THE TRANSPORT OF REE'S IN HYDROTHERMAL FLUIDS

Humphries (1984) stated that transport of REE's in hydrothermal fluids for any great distance is probably accomplished by the formation of carbonate, fluoride, or sulfate complexes. In addition, experimental studies cited by Humphries (1984) indicate that for most REE's, RE(Cl,F)2+ has been identified as the most stable complex in acidic solutions rich in Cl- or F- ions, but RE(Cl,F)3 is most stable for La. The presence of fluoride significantly increases the mobility of the REE's.

For the early stage of disseminated monazite in ores hosted by H8 marble, REE's clearly were associated with only P anions, and therefore the acidic REE hydrothermal solutions were enriched in P ions only. In the case of the early-stage precipitation of disseminated bastnaesite, F was introduced, so that the most stable complexes could have been RE(Cl,F)2+ and RE(Cl,F)3.

Monazite and bastnaesite in banded ores could also have resulted from precipitation of complex ions of carbonates or fluorides from the hydrothermal fluids; the bands of monazite and bastnaesite are probably unrelated to the associated fluorite bands, which are in general later in the paragenetic sequence of the assemblage. The early stage of disseminated REE mineralization and the peak period of REE mineralization in banded ores were apparently not related to any alkali cations because aegirine augite and alkali amphiboles, if present, always belong to later stages in the paragenetic sequence of those assemblages.

Although casual inspection of the complex Bayan Obo deposit may lead to the conclusion that the associations of alkali amphiboles, aegirine augite, and REE minerals are related to a carbonatite magma, the experimental data from Humphries (1984) and the younger ages of the amphiboles and augite with respect to associated REE minerals refute this. Multiple episodes of REE mineralization span a period of about 150 million years at Bayan Obo (Wang and others, 1994), and, therefore, invoking a carbonatite source for REE's would also require evidence to indicate carbonatite magmatic activity lasting 150 million years. Bayan Obo, thus, is a prime and valuable example of the importance of paragenetic sequence, rather than mineral assemblages, in discerning the history and origin of a deposit, as well as the fact that associations of REE minerals, alkali amphiboles, and aegirine augite need not invoke a carbonatite source.

We were unable to obtain any information on the nature of the fluids from fluid inclusions. We found no fluid inclusions in any of the REE minerals we have studied. Fluid inclusions are most abundant in late-stage barite, which is essentially unrelated to specific episodes of REE mineralization. Minute fluid inclusions are also present in fluorite. Limited data on such inclusions were published by the Institute of Geochemistry (1988).

AGE OR AGES OF EPISODES OF MINERALIZATION

Bayan Obo is uniquely informative with respect to precise ages of REE mineralization and relative sequence of Fe and Nb mineralization. Episodes of REE mineralization began about 555 Ma and lasted until about 400 Ma, a period of about 150 m.y. Fe mineralization began about 430 Ma and lasted until after about 343 Ma. Bayan Obo is also unique in that the complex metamorphic and metasomatic histories can be documented by minimum mineral ages of the alkali amphiboles present, which cover a range from about 1.26 Ga to about 300 Ma.

LITHOLOGY AND AGE OF RESERVOIR ROCKS FAVORABLE FOR HOSTING THE ORES

Although the H8 marble is not the only rock type that hosts Bayan Obo ores, it is clearly the dominant host rock. As marble, the reservoir host rock was particularly reactive with acid hydrothermal solutions, producing CO2 when the carbonate was dissolved by the acid solutions. Contemporaneously, the metals precipitated from the solution because of cooling and change of pH when the acid solution was neutralized. Clearly, carbonate host rocks rank as the most favorable reservoir host rocks for these ores.

Besides lithology, permeability of the host rock is also an important property to be considered. Textural evidence from Bayan Obo ores suggests that replacement of the host rock minerals was predominantly along grain boundaries and along microfractures. When extensive metasomatic replacement occurred, the host rock became completely replaced, as shown by some of the banded ores.

The age of the host rock is apparently also important. H8 is of Middle Proterozoic age, which existed prior to Caledonian episodes of mineralization. Any carbonate host rocks younger than the Caledonian would not have been mineralized. Hence, Proterozoic carbonate host rocks overlying Archean and Early Proterozoic source rocks, in conjunction with younger tectonic processes such as subduction or rifting, represent a good combination for the formation of these ore deposits.

PRESERVATION OF THE ORES (CAP ROCKS)

Another favorable condition for the formation of the Bayan Obo ores results from the presence of H9 biotite and albitized biotite schist, which is pelitic and less permeable than the underlying H8 host rocks. Although not perfectly impermeable, the H9 biotite schist acted as a cap rock and slowed down or stopped the escape of the incoming acid hydrothermal solutions. The ores precipitated by the solutions were therefore confined and preserved.

BAYAN OBO AND THE PORPHYRY Cu-Mo ORE DEPOSITS OF PERU AND CHILE

In order to gain further insight regarding giant and super giant ore deposits of hydrothermal origin, it may be useful to compare the Bayan Obo deposit with the Andean giant and super giant porphyry Cu-Mo ore deposits of South America from Peru to Chile, although they are different ore types. Whether the key factor of giant ore deposits is the tectonic setting that sustains the long duration of mineralization should be carefully examined and tested.

Using the National Academy of Science size classification of ore deposits, Clark (1993) listed two Andean behemothian (>31 million metric tons of Cu) porphyry Cu ore deposits (El Tiniente and Chuquicomata) and seven super giant (>10 million metric tons of Cu) porphyry Cu ore deposits (El Abra, Chuqui Norte, Los Pelambres-El Pachon, Rio Blanco-Los Bronces, La Escondita, Collahuasi-Rosario, and Mansa Mina). However, from the economic geology point of view, the best studied Andean porphyry Cu-Mo deposit is the giant El Salvador, which is used as a standard for this type of ore deposit (Gustafson and Hunt, 1975). Of particular interest is the long period over which these deposits formed, which, on the basis of the ages of the porphyries, lasted approximately 95 m.y. (100-5 Ma) (Clark, 1993, fig. 10).

On the basis of the study of El Salvador, most investigators favor the interpretation that the subvolcanic porphyries with which these giant, super giant, and behemothian Cu-Mo ore deposits are associated were related to the subduction of the Nazca (Farallon) oceanic plate (Clark, 1993, fig. 12). Hydrothermal solutions that deposited Cu-sulfide veins were directly related to the mantle-derived melt that resulted in crystallization of the porphyry stocks close to the surface. Clark, using a study by Pardo-Casas and Molnar of 1987 (cited in Clark, 1993), summarized the timing of mineralization with respect to the displacement of the Nazca (Farallon) plate relative to South America as follows: (a) before and during early Eocene (62-52 Ma), porphyry copper emplacement in southern Peru and northernmost Chile; (b) before and during late Eocene and early Oligocene (42-31 Ma), emplacement in northern Chile; and (c) since 10.59 Ma, emplacement of central Chilean porphyry Cu deposits. Thus, regionally the formation of Andean porphyry ore deposits migrated from north to south within about 50 million years.

The Andean porphyry Cu-Mo ore deposits are no doubt of hydrothermal origin because the principal ore metals occur mostly in two of three types of veins in the marginal zones of porphyries, according to Gustafson and Hunt (1975). The most important and the earliest Cu mineralization at El Salvador formed the A veins disseminated with chalcopyrite-bornite and locally, traces of molybdenite. B veins crosscut A veins and were clearly later. B veins characteristically contain molybdenite-chalcopyrite and traces of bornite in some of the veins. D veins, poor in quartz, crosscut both A and B veins and are the youngest veins. Pyrite is usually predominant in D veins, and chalcopyrite, bornite, enargite, tennantite, sphalerite, and galena are also present. Minor molybdenite and many other sulfides occur locally.

Gustafson and Hunt (1975) concluded that, based on isotopic age data, the period of mineralization in the veins and associated porphyry is essentially less than a million years, not resolvable by the K-Ar method of dating available at the time of their study. Redating the veins and porphyries using the 40Ar/39Ar and Re-Os methods may provide additional information about the mineralization history at El Salvador. The porphyry stocks that intruded the volcanic-covered terrain required time to cool before fracturing could occur to provide openings for the deposition of Cu ore in veins. If one assumes the cooling rate of a porphyry stock to be 50-10°C/m.y. and if the original melt-mush was at a temperature above 800°C, it would require several million years to cool before fracturing could occur. Therefore, dating the porphyries with the 40Ar/39Ar method or with Rb/Sr mineral isochrons and dating the age of the veins directly using the Re-Os method on Cu-sulfides and molybdenite may provide new insights about the duration of mineralization at El Salvador.

The basic question of interest is whether the hydrothermal solutions necessarily are related to the generation of the porphyry melt by subduction of the Nazca (Farallon) plate. Subduction-related heating could easily have generated the porphyry melt and the hydrothermal solutions enriched in sulfides at essentially the same time, or the hydrothermal solutions may have been generated several million years later than and independent of the melt.

Bayan Obo appears to have some aspects in common with the porphyry Cu-Mo ore deposits of the Andes: (1) they both lie along the margin of a craton affected by subduction of an oceanic plate; (2) they both have a long duration of formation of ore, covering a span of more than 50 m.y.; and (3) they are both of hydrothermal origin. The principal differences are (1) they are different deposit types; (2) Bayan Obo episodes of mineralization were restricted to a single deposit, whereas the Andean porphyry Cu-Mo deposits, representing a span of more than 90 m.y., consist of many separate giant and super giant deposits of similar types; and (3) evidence for Caledonian subduction in the Bayan Obo area is restricted to an 850-km trace, whereas according to Clark (1993), the location of subduction of the Nazca (Farallon) plate under the Andean crust moved from north under Peru, to south under southern Chile, to east under northern Chile. This movement would account for the sequence and locations of the porphyry Cu-Mo deposits of the Andean belt.

In the case of Bayan Obo, there is no evidence of any igneous melt related to the hydrothermal solutions that deposited the REE's, Fe, and Nb of that deposit. The key lies in the source area, where the subduction-related heat generation caused scavenging of ore metals from the lower crust and enrichment of ore metals in the hydrothermal solutions. Within the mine region (16 km by 2-3 km) at Bayan Obo, multiple episodes of REE mineralization, hypothetically generated by Caledonian subduction, lasted more than 150 m.y. With regard to the Andean porphyry Cu-Mo deposits, episodic subduction (from before 60 Ma to about 5 Ma) activated the mantle-crust source region to produce the world's largest porphyry Cu-Mo ore deposits.

One may also wish to compare Bayan Obo and the Andean porphyry Cu-Mo ore deposits with the Sn granite ore deposits of Bolivia, central South America, or other areas of the world, to see if activities in certain tectonic settings, such as rifting or subduction, can cause repeated generation of either ore-related igneous melts or ore-related hydrothermal solutions. Similarities of types of ore deposits are not necessarily required as a key to the origin of giant ore deposits in favorable tectonic settings.